Robotic Tactile Perception of Object Properties: A Review (1711.03810v1)

Abstract: Touch sensing can help robots understand their sur- rounding environment, and in particular the objects they interact with. To this end, roboticists have, in the last few decades, developed several tactile sensing solutions, extensively reported in the literature. Research into interpreting the conveyed tactile information has also started to attract increasing attention in recent years. However, a comprehensive study on this topic is yet to be reported. In an effort to collect and summarize the major scientific achievements in the area, this survey extensively reviews current trends in robot tactile perception of object properties. Available tactile sensing technologies are briefly presented before an extensive review on tactile recognition of object properties. The object properties that are targeted by this review are shape, surface material and object pose. The role of touch sensing in combination with other sensing sources is also discussed. In this review, open issues are identified and future directions for applying tactile sensing in different tasks are suggested.

• The paper provides an exhaustive review of tactile sensing technologies and fusion methods for assessing object materials, shapes, and poses.
• It details methodologies using classifiers, neural networks, and SLAM-inspired tracking to boost the accuracy of tactile perception.
• The review underscores the promise of integrating multimodal sensory data to advance human-like manipulation in unstructured environments.

Insightful Overview of "Robotic Tactile Perception of Object Properties: A Review"

The paper entitled "Robotic Tactile Perception of Object Properties: A Review" provides an exhaustive survey of the developments and scientific achievements in the domain of robotic tactile systems. It focuses on the comprehensive understanding and interpretation of tactile data for assessing object properties, which is integral to enhancing the autonomy and decision-making of robots in unstructured environments.

Tactile Sensing Technologies and Modalities

The review commences with an examination of tactile sensing technologies, categorizing them based on sensor types and their analogous functions to human touch, such as single-point contact sensors, high spatial resolution tactile arrays, and large-area tactile sensors. Each category offers unique advantages, suitability, and limitations for different tactile tasks and applications, from confirming contact and measuring force or vibrations to providing detailed touch sensing akin to human fingertips.

Object Property Recognition

The review thoroughly discusses methodologies for object property recognition through tactile sensing. The three main areas of focus are the recognition of object materials, shapes, and pose estimation.

1. Material Recognition: The paper reviews multiple tactile-based methods for material recognition, leveraging surface texture information through friction, acoustic signals, and micro-pattern arrays, in addition to assessing mechanical impedance measures. Emphasizing algorithms for interpreting these tactile modalities, the authors refer to techniques like Bayes classifiers, neural networks, and support vector machines (SVM) for achieving robust material discrimination.
2. Shape Perception: Distinctions are made between local and global shape recognition. The paper reflects on using vision-inspired descriptors to treat tactile data as images, employing PCA-based features, and employing self-organizing feature learning to enhance the robustness and accuracy of shape recognition. For global shape perception, methods under investigation utilize contact points, integrate tactile patterns, or combine both spatial distributions and local features.
3. Pose Estimation: The manuscript surveys the state-of-the-art in using tactile sensing, either solely or in conjunction with visual sensors, for precise object pose estimation. Discussions involve tracking methodologies reminiscent of SLAM algorithms in mobile robotics, alongside recent advances in integrating visual-tactile data to address limitations of purely visual systems due to occlusions during manipulation.

Contributions to Sensor Fusion

The utility of tactile sensors in sensor fusion, particularly with vision, is detailed, emphasizing their role beyond simple contact confirmation. Tactile sensors supplement visual data to enhance model construction, assist in feature extraction, and provide intricate local information. There is a focus on frameworks that leverage domains such as Bayesian inference and probabilistic modeling to fuse tactile inputs with other sensory streams, furthering applications in object recognition, manipulation, and grasping tasks.

Implications and Future Directions

The paper’s survey of tactile perception indicates significant implications for the ongoing development of robotic systems—specifically, developments aiming to achieve human-like manipulation capabilities. The convergence of tactile data with other sensory modalities suggests future research should focus on the seamless integration of multimodal data and the exploration of unsupervised learning approaches to improve the adaptability and generalization of tactile perception frameworks. Additionally, optimizing the balance between sensor resolution, cost, and computational efficiency remains a critical challenge.

Overall, this extensive review is invaluable for researchers aiming to extend tactile sensing technologies’ practical applications, providing a robust foundational understanding and identifying key challenges for future exploration in the field.